Skip to main content

Full text of "Github The Chymera OPR 9c 6a 4b 9ce 5659f 03b 3e 76d 81ef 3f 594e 0814460f"

See other formats

Published for OPR according to 

Concepts and Guidelines for an 
Open Publication Revision" 


Horea Christian 12 

Scientific data is being produced at an ever in- 
creasing rate, challenging our means of com- 
municating findings. Currently, journals act 
as filters, and use peer review of context-data- 
evaluation-and-discussion monoliths to make 
a go/no-go choice regarding publication. We 
propose a non-binary model which harnesses 
the power of open collaboration, immediate 
publication, and transparent revision. We 
name this model, accordingly, Open Publi- 
cation Revision (OPR). We believe that the 
worthiness of findings is never static, and can 
only be interpreted in light of their ability 
to support further developments. Our model 
separates publication from revision and from 
quality assessment, and gives every finding 
the opportunity to make its way into highly 
regarded ensembles of knowledge. 

1 Background 

Peer review is a widely acclaimed procedure for qual- 
ity control, the principle of which has circulated in 
human societies for millennia [22]. Its rationale is 
both solid and simple: Only people familiar with a 
certain part of the knowledge network can decide 
whether and how a new part fits in. 

The exact implementation of peer review in the 
context of publishing has changed in accordance 
with the technological possibilities of the time [22] , 
the most recent notable changes including online 
publishing and review, and the more controversial 
open access publishing [24, 17] and open peer review. 
While open access seems to have gained a foothold in 

the scientific community (notably demonstrated by 
the recent success of the Public Library of Science 
journals), the same cannot be said of open peer 

Many opinion and research articles concern 
themselves with the benefits[14, 15] and the 
applicability [25, 26] of open peer review. It should 
be noted, however, that this is but the most promi- 
nent of many incentives to change peer review. A 
landmark of the discussion of modern publishing was 
Nature's 2006 series on the trial and debate of peer 
review [4]. The debate illustrates that the scientific 
community is bursting with ideas regarding the fu- 
ture of publishing. Some of the articles also show 
how novel systems have already been implemented 
[20, 21, 11], and how one in particular, arXiv[l], has 
even become a standard in a number of fields. 

The article series shows that seasoned members of 
the scientific publishing community are vividly aware 
of a great many shortcomings of current knowledge 
mediation in general and peer review in particular. 
Such disappointments, complemented by many oth- 
ers, are also shared by less established authors [15]. 

2 Issues 

From the vast pool of opinions in the relevant lit- 
erature, we have selected a number of knowledge 
mediation issues which we consider most pressing, 
and to which we presume to offer pertinent solutions. 
For ease of overview we use "peer review" to refer 
to the predominant current implementation thereof 
(including such aspects as there being 3 reviewers 
per article). 


Licensed under the Creative Commons Attribution-SharcAlikc 3.0 Unported License. 

page 1 of 5 

Published for OPR according to 

1. Peer review is time-consuming. This can cause 
excessive delay of rapidly outdated findings [20] , 
or conflicts between authors publishing similar 

2. Peer review commonly imposes a priori gauged 
"interest" as a filter to publishing [6]. This 
may encourage the formation of dogmata [5]. 
Anecdotally, the unreliability of such foresight 
has lead to the rejection of Nobel-level articles 

3. Reviewers are expected to contribute without 
receiving compensation for their work - but risk 
losing face if consistently refusing. This lack 
of positive incentive for review makes it a ma- 
jor bottleneck in publishing, and a burden for 
researchers [11]. 

4. Peer review creates hierarchic value for journals 
[10, 27], rather than for articles. There is little 
incentive for peer review coordinators to make 
an article as good as it possibly can be at a 
great expense of effort; the incentive is rather 
to select the probably most citable articles with 
minimum effort. 

5. Peer review is left to those with the least time 
and hands-on experience. Practical expertise 
of technicians and plentiful time resources of 
students are utterly disregarded in the process 


6. The peer review of their own work is often in- 
transparent to researchers, but even more so to 
the public [7]. 

3 Methods 
3.1 Git 

The main method we use to tackle these issues is elec- 
tronic revision control - a concept most widespread 
in software development. This powerful system al- 
lows authors of scientific articles to benefit from the 
same features which enable programmers around the 
world to collaboratively and asynchronously build 
functioning code online. The history of revision 
control exceeds the scope of this article; though it 
should be noted that more basic forms of revision 
control (such as book editions) already have a strong 
tradition in knowledge management. 

Of the modern implementations of revision con- 
trol we have chosen to use Git [23] - owing to its 
distributed and decentralized nature, and support 
for non-linear development. 

For online publishing we encourage using github[8]. 
Owing to the nature of Git, however, users will be 
able to remain integrated in the network regardless 
where they choose to host. 

3.2 LaTeX 

Our method of choice for keeping information orga- 
nized within an article (while supporting easy human 
readability and a pleasurable visual experience) is 
the LaTeX [13] document markup language. LaTeX 
is already a standard for a great number of publish- 
ers; and implementing it would potentially facilitate 
collaboration with journals. 

4 Results 

4.1 Overview 

In brief, we believe the issues we address would be 
resolved in the following manner: 

1. Immediate publication allows anyone who is 
ready to sacrifice polish for rapid access to data 
the means to do so. Articles would receive revi- 
sion time proportional to how many people they 
motivate; and authors publishing similar stud- 
ies would rather collaborate than try to obscure 
each other. 

2. By removing any filter and assigning the task of 
evoking interest to the evolving article; articles 
would fairly compete for citation and furthering 
of their findings. 

3. Revisions are directly attributed to their au- 
thors, so there is a strong incentive to invest 
time in foreign work (authors of significant re- 
visions become co-authors of future versions). 
Researchers can choose their own personal mix 
of review and primary publishing work, and 
receive true effort-proportional recognition. 

4. Value is directly assigned to articles, and the 
chief incentive for all who revise is to make their 
work as useful as possible. 

5. Everyone is able to revise articles and contribute 
their strongest personal expertise to the knowl- 
edge network. 

6. OPR as a system would be completely transpar- 
ent - to members of the public, journalists, and 
policy makers. 


Licensed under the Creative Commons Attribution-Share Alike 3.0 Unported License. 

page 2 of 5 

Published for OPR according to 

4.2 Workflow 

The aforementioned emergent properties owe to the 
following workflow: 

• One or multiple initial authors publish an article 
whenever they feel they should. 

• Researchers interested in the topic find the arti- 
cle (via the web, or via advertisement from the 
original authors), and read it. 

• Depending on how reliable they find the work 
and interesting the finding, other researches may 
cite, further recommend, or fork the article into 
their own repository and revise it. 

• Authors of revisions would be interested in fur- 
thering the work which is now also their own, 
and would again propagate the article (reiterat- 
ing the workflow at the second point). 

• The original authors may choose to adopt none, 
one, or multiple of the revisions and merge them 
into their repository. 

• Authors of several articles may also merge their 
work to increase its scope and visibility. 

It is worth noting that the above process, though 
intricate and non-linear, is expeditious. Contingent 
on the number of interested researchers, an article 
could undergo manifold circles of revision, forking, 
and merger in the months usually required for vali- 
dation through peer review. 

4.3 Copyright 

We suggest licensing any article intended for OPR 
under a "Creative Commons Attribution-ShareAlike 
Unported License" (preferably of the most recent 
version). This allows anyone - including journal edi- 
tors, educators, and policy makers - to build upon an 
article, and ensures both the continuity of attribu- 
tion to the researcher (s), and the persistent freedom 
of the information. 

4.4 Authorship 

In Git each revision of every single line is attributed 
to its specific author. Article authorship is thus 
defined clearer in OPR than in any other means of 
research publication. We do realize, though, that 
this very semantic notion of authorship is difficult 
to reconcile with the currently prevailing concept in 
scientific publishing. We are also aware that article 
authorship is not identical with research authorship. 

We propose an explicit mention of authors (much 
like in most scientific journals) beneath the title. 
On the occasion of each revision, the author thereof 
should append his name, and those of any who may 
have helped him (in whatever order he deems fair), 
to the end of that list. Names should, of course, not 
be repeated if they are already on the authors' list. 

Whenever two or more articles are merged, the 
order in which their authors' lists are concatenated 
is left to the discretion of the person performing the 

Independently of the initial number of authors we 
encourage using the editorial "we" in the wording of 

4.5 Article Structuring 

We propose structuring articles along the following 

• Background - detailing the context of what- 
ever the article discusses. 

• Issues - detailing precisely and explicitly what 
the article presumes to tackle. 

• Methods - detailing the (technical) means by 
which the issues are to be tackled. 

• Results - detailing how usage of the methods 
materializes as far as the author was able to 

Subsections of these items are left to the author's 
discretion, but we do insist on everyone strictly ad- 
hering by these four overhead categories. Rigorous 
implementation of this model is expected to assure 
homogeneity, clarity, and relative ease of performing 
more complex operations such as merging, forking, 
or text data mining. 

We explicitly and strongly discourage using the 
popular "Discussion" category, as we expect it to 
provide a playground for highly speculative debate. 
We encourage brevity in each category. 

In the hope of supporting articles from the hu- 
manities in OPR, thought experiments may also 
be considered valid sources of information for the 
Results section. 

4.6 File Management 

The LaTeX template which we propose for all OPR 
texts is the one demonstrated by this article. We 
believe it is of utmost importance that we maintain 
a consistent visual appearance. 

The template should be copied and pasted, and 
the contents replaced. Original content is written 


Licensed under the Creative Commons Attribution-ShareAlike 3.0 Unported License. 

page 3 of 5 

Published for OPR according to 

in article.tex and the abstract in README. Ref- 
erences are listed under bib. bib. Metadata such as 
comments on the template should be left untouched, 
and vital markup modifications should be referred 
to upstream. 

In order to better facilitate prospective data min- 
ing and semantic authorship every sentence or sub- 
sentence (for instance following the usage of a semi- 
colon), should be placed on a new line. The LaTeX 
interpreter conveniently does not introduce a break 
in this context. 

The Git repository name for OPR articles should 
begin with " OPR." followed by the title, contracted 
as the author sees fit. The reasons for contraction 
are mostly visual, since Git hosting services tailor 
their websites to short repository names. 

Exported PDF files and figures should only be 
pushed to a repository for milestone versions (they 
are explicitly excluded via .gitignore). It should 
be noted that Git keeps track of all changes made 
to files, and binaries lead to very rapid bloating of 
repositories. Image files should be tailored to the 
minimum adequate size for quality PDF printing - 
85um/dot (300 DPI). Overall, the PDF files in a 
repository should only be considered a convenience 
for readers. 

All citations of OPR articles should link to 
a specific revision via its hash (and preferably 
give a link). In case readers should decide 
to disseminate loose PDF files, they should ap- 
pend the version hash (and repository) to the file 
name. Example: github-repository-ArticleName- 

4.7 Supplementary Information 

OPR only makes sense if contributors can carefully 
scrutinize the data and accurately retrace its process- 
ing and analysis. We therefore encourage authors to 
publish their data as well as their post-processing and 
analysis scripts online, and license them as liberally 
as possible (ideally also under a "Creative Commons 
Attribution-ShareAlike Unported License"). 

Authors should version post-processing and analy- 
sis scripts with Git - so that contribution to those is 
also possible. We encourage the use of purpose-built 
scripts in a transparent language (such as Python). 
For the sake of transparency and accessibility, we 
would recommend FOSS alternatives to proprietary 
packages, for instance: 

• Scipy-Numpy-Matplotlib[16] over MATLAB 

• R[9] or PSPP[2] over SPSS 

• PsychoPy[18, 19] over E-Prime and Presenta- 

If given a choice, OPR materials should cite open 
access articles rather than restricted access articles 
of similar quality. Doing so maximizes the extent to 
which potential contributors and readers can assess 
whether or not assumptions used in the article are 

4.8 Implementation 

We acknowledge that our publishing system does not 
provide any alleged certificate of quality. While we 
expect OPR articles to become superior in quality 
to comparable peer-reviewed static articles, we are 
aware that the burden of proof still rests with us. It 
would currently be unwise for researchers to commit 
articles with high-profile potential to OPR. There- 
fore, we recommend to use OPR first and foremost 
to publish work which journals are reticent to accept, 
such as: 

• Negative findings 

• Discontinued work 

• Ongoing work 

• Informal studies conducted in researchers' free 

By acting as a hub for such findings OPR can already 
and immediately provide a valuable service to the 
scientific community - since work of this type would 
otherwise be lost. In fact, revision is most needed 
for precisely these types of articles; and improving 
them is an excellent challenge for our system to face. 
If OPR can make a name by adding value to less 
promising articles, then indeed it is more laudable 
than any publishing approach which derives value 
for itself from more promising ones. 


1 Interdisciplinary Centre for Neurosciences (IZN), Univer- 
sity of Heidelberg, INF 364 69120 Heidelberg, Germany. 

2 Department of Experimental Psychology, Oxford Univer- 
sity, Oxford, UK 


[1] General information about arxiv. URL: http://arxiv. 

[2] Statistical analysis of sampled data. URL: http://www. 
gnu . org/sof tware/pspp/. 

[3] Coping with peer rejection. Nature, 425(6959):645, Oct 
2003. URL: http : //dx . doi . org/10 . 1038/425645a, doi : 


Licensed under the Creative Commons Attribution-ShareAlike 3.0 Unported License. 

page 4 of 5 

Published for OPR according to 

[4] 2006. URL: 

peerreview/debate/nature05535 . html, doi : doi : 

10 . 1038/nature05535. 

[5] Richard Akerman. Technical solutions: Evolving peer re- 
view for the internet. Nature, 2006. URL: http : //dx . doi . 
org/10 . 1038/nature04997, doi : 10 . 1038/nature04997. 

[6] Theodora Bloom. Online frontiers of the peer-reviewed 
literature. Nature, 2006. URL: 
1038/nature05030, doi : 10 . 1038/nature05030. 

[7] Tracey Brown. I don't know what to believe. Nature, 

2006. URL:, 
doi : 10 . 1038/nature04998. 

[8] Inc. GitHub. Github - social coding. URL: https:// 
github. com/. 

[9] Ross Ihaka and Robert Gentleman. Free software envi- 
ronment for statistical computing and graphics. URL: 
http: //www.r-project .org/. 

[10] Charles Jennings. Quality and value: The true purpose 
of peer review. Nature, 2006. URL: 
10 . 1038/nature05032, doi : 10 . 1038/nature05032. 

[11] Thomas Koop. An open, two-stage peer-review jour- 
nal. Nature, 2006. URL: 
nature04988, doi : 10 . 1038/nature04988. 

[12] Debomoy Lahiri. Perspective: The case for group re- 
view. Nature, 2006. URL: 
nature05033, doi : 10 . 1038/nature05033. 

[13] Leslie Lamport. Latex a document preparation system. 
URL: http: //www. latex-project .org/. 

[14] Jeffrey T. Leek, Margaret A. Taub, and Fernando J. 
Pineda. Cooperation between referees and authors in- 
creases peer review accuracy. PLoS One, 6(ll):e26895, 
2011. URL: 
pone . 0026895, doi : 10 . 1371/ journal . pone . 0026895. 

[15] Gaell Mainguy, Mohammad R. Motamedi, and Daniel 
Mietchen. Peer review-the newcomers' perspective. 
PLoS Biol, 3(9):e326, Sep 2005. URL: http:// 
dx .doi . org/10 . 1371/ journal .pbio . 0030326, doi : 10 . 
1371/ j ournal .pbio . 0030326. 

[16] Travis Oliphant. Fundamental package for scientific com- 
puting with python. URL: 

[17] Michael Parker. The ethics of open access publishing. 
BMC Med Ethics, 14(1):16, Mar 2013. URL: http:// 
dx. doi. org/10. 1186/1472-6939-14-16, doi: 10. 1186/ 

[18] Jonathan Peirce. Open-source application to allow the 
presentation of stimuli and collection of data for neu- 
roscience, psychology, and psychophysics. URL: http: 
//www . psychopy . org/. 

[19] Jonathan W. Peirce. Psychopy-psychophysics software 
in python. J Neurosci Methods, 162(l-2):8-13, May 

2007. URL: http://dx.doi.Org/10.1016/j.jneumeth. 
2006 . 11 . 017, doi : 10 . 1016/ j . jneumeth . 2006 . 11 . 017. 

[20] Brenda Riley. Trusting data's quality. Nature, 2006. 
URL:, doi: 
10 . 1038/nature04993. 

[21] Erik Sandewall. Opening up the process. Nature, 2006. 
URL:, doi: 
10 . 1038/nature04994. 

[22] Ray Spier. The history of the peer-review process. Trends 
Biotechnol, 20(8):357-358, Aug 2002. 

[23] Linus Torvalds. git -distributed-is-the-new-centralized. 

[24] Richard Van Noorden. Open access: The true cost 
of science publishing. Nature, 495(7442) :426-429, Mar 
2013. URL: http : //dx . doi . org/10 . 1038/495426a, doi : 

[25] S. van Rooyen, F. Godlee, S. Evans, N. Black, and 
R. Smith. Effect of open peer review on quality of reviews 
and on reviewers' recommendations: a randomised trial. 
BMJ, 318(7175) :23-27, Jan 1999. 

[26] Susan van Rooyen, Tony Delamothe, and Stephen J W. 
Evans. Effect on peer review of telling reviewers that their 
signed reviews might be posted on the web: randomised 
controlled trial. BMJ, 341x5729, 2010. 

[27] Elizabeth Wager. What is it for? analysing the purpose 
of peer review. Nature, 2006. URL: 
10 . 1038/nature04990, doi : 10 . 1038/nature04990. 


Licensed under the Creative Commons Attribution-SharcAlikc 3.0 Unportcd License. 

page 5 of 5